To　solve　the　problem　of　the　location　of　the　fault　point　of　single-pole-to-ground　faults　in　the　transmission　lines　of　MMC-HVDC　systems,　this　paper　designs　a　fault　location　system　based　on　support　vector　machine　(SVM).　The　waveform　of　the　traveling　wave　after　the　fault　occurs　is　collected　as　a　feature,　and　the　regression　mechanism　of　the　SVM　is　utilized　to　achieve　fault　location.　Because　it　is　very　difficult　to　locate　high-resistance　ground　faults,　this　paper　first　analyzes　the　waveform　characteristics　of　high-resistance　ground　faults.　Next,　three　steps　are　proposed　to　reduce　the　influence　of　grounding　resistance　on　fault　location.　These　steps　include　using　the　active　pulse　waveform　as　a　new　feature,　classifying　the　samples　according　to　ground　resistance　values　before　training　regression　models,　and　a　method　for　adaptively　extracting　fault　distance　features　is　proposed.　Finally,　a　complete　location　system　design　is　proposed,　and　its　workflow　is　illustrated.　After　the　simulation　test,　the　proposed　location　system　only　needs　to　obtain　a　single-ended　fault　voltage　waveform　at　a　fault　recording　frequency　of　20　kHz　to　achieve　an　accurate　location　of　single-pole-to-ground　faults　for　different　values　of　grounding　resistance.　©　2013　IEEE.